This paper presents an explainable AI (XAI) approach for Alzheimer's disease (AD) diagnosis using deep transfer learning. The study proposes an ensemble model combining VGG16, VGG19, DenseNet169, and DenseNet201 to achieve high accuracy in AD classification. The proposed model, which integrates XAI techniques such as saliency maps and grad-CAM, achieves an impressive accuracy of 96%. The ensemble models, Ensemble-1 (VGG16 and VGG19) and Ensemble-2 (DenseNet169 and DenseNet201), demonstrate superior performance with accuracy, precision, recall, and F1 scores reaching up to 95%. The XAI techniques enhance model interpretability, providing clinicians with visual insights into the neural regions influencing diagnostic decisions. The study also evaluates the performance of various models on a comprehensive dataset, highlighting the effectiveness of deep transfer learning and XAI in AD diagnosis. The results show that the proposed model outperforms existing methods in terms of accuracy and interpretability, offering a promising solution for more transparent and clinically relevant AI models in healthcare. The integration of XAI techniques not only improves diagnostic accuracy but also supports the understanding of the decision-making process in AI models. The study contributes to the field of AD diagnosis by developing a robust deep transfer learning ensemble model and introducing a novel diagnostic model with high accuracy and interpretability. Future work includes further exploration of deep transfer learning ensembles, refinement of XAI techniques, and evaluation on larger datasets to enhance the model's generalizability and robustness.This paper presents an explainable AI (XAI) approach for Alzheimer's disease (AD) diagnosis using deep transfer learning. The study proposes an ensemble model combining VGG16, VGG19, DenseNet169, and DenseNet201 to achieve high accuracy in AD classification. The proposed model, which integrates XAI techniques such as saliency maps and grad-CAM, achieves an impressive accuracy of 96%. The ensemble models, Ensemble-1 (VGG16 and VGG19) and Ensemble-2 (DenseNet169 and DenseNet201), demonstrate superior performance with accuracy, precision, recall, and F1 scores reaching up to 95%. The XAI techniques enhance model interpretability, providing clinicians with visual insights into the neural regions influencing diagnostic decisions. The study also evaluates the performance of various models on a comprehensive dataset, highlighting the effectiveness of deep transfer learning and XAI in AD diagnosis. The results show that the proposed model outperforms existing methods in terms of accuracy and interpretability, offering a promising solution for more transparent and clinically relevant AI models in healthcare. The integration of XAI techniques not only improves diagnostic accuracy but also supports the understanding of the decision-making process in AI models. The study contributes to the field of AD diagnosis by developing a robust deep transfer learning ensemble model and introducing a novel diagnostic model with high accuracy and interpretability. Future work includes further exploration of deep transfer learning ensembles, refinement of XAI techniques, and evaluation on larger datasets to enhance the model's generalizability and robustness.